Data repositories in which cases are related to subcases are identified as hierarchical. This course covers the representation schemes of hierarchies and algorithms that enable analysis of hierarchical data, as well as provides opportunities to apply several methods of analysis.

Associate Professor at Arizona State University in the School of Computing, Informatics & Decision Systems Engineering and Director of the Center for Accelerating Operational Efficiency School of Computing, Informatics & Decision Systems Engineering

K. Selcuk Candan

Professor of Computer Science and Engineering Director of ASU’s Center for Assured and Scalable Data Engineering (CASCADE)

Okay, let's summarize.

So in this video,

we learned about ways to compare time series.

We had said that there are,

overall, sort of two classes of approaches.

We can either basically cause the synchronized approaches,

like Euclidean distance and correlation,

where we are assuming that

the two time series are sort of on

the same time scale and they are working on the same time frame.

These are often relatively cheap to compute because

I am essentially have to consider only the corresponding entries in both time series,

but these algorithms don't account for things like shifts,

stretches, and other time frame misalignments.

Asynchronous approaches, however, they don't necessarily need to make the assumption

that the time frames or the time series are aligned or they have the same speed.

So, they can account for misalignments in

the shift or in terms of stretch between the given time series.

This, however, are usually quite expensive to compute.

We have learned edit distance and dynamic time warping,

and both of them are quadratic in cost.

That is, if one time series is of length N and the other time series of length M,

the cost of the competition is N times M. So that's the amount of work that I have to do.

So, the solutions to this problem include either constraining the search space or

to first obtain a reduced representation of the time series

and then compute the distance or similarity on that reduced representation.